Facilitation contributes to Mediterranean woody ... - Wiley Online Library

Research

Facilitation contributes to Mediterranean woody plant diversity but does not shape the diversity–productivity relationship along aridity gradients Pedro J. Rey, Julio M. Alc
antara, Antonio J. Manzaneda and Alfonso M. S
a nchez-Lafuente Department of Biolog
ıa Animal, Biolog
ıa Vegetal y Ecolog
ıa, Universidad de Ja
en, E-23071 Ja
en, Spain

Summary Author for correspondence: Pedro J. Rey Tel: +34 953 212145 Email: [email protected] Received: 11 December 2015 Accepted: 29 January 2016

New Phytologist (2016) 211: 464–476 doi: 10.1111/nph.13916

Key words: competition–facilitation balance, drylands, humped-back diversity– productivity relationship, nurse, plant positive interactions, relative interaction index, species richness, stress-gradient hypothesis.


The diversity–productivity relationship (humped-back model (HBM)) and the stressgradient (SGH) hypotheses may be connected when productivity is limited primarily by aridity. We analytically connect both hypotheses and assess the contribution of facilitation to woody plant richness along the aridity gradient of the Western Mediterranean floristic region.
We monitored regeneration niches of woody plants, obtaining rarefied species richness and plant relative interaction indices in 54 forests and scrublands in a 1750-km geographical range across Spain, Morocco and the Canary Islands.
We verified the monotonic increase in facilitation with aridity postulated by SGH and the humped-shape pattern of species richness expected from HBM, which became manifest after expanding the aridity gradient or crossing vegetation types. Along the gradient, interaction balance turned into facilitation earlier in forest than in scrublands. The effects of aridity and interaction balance on species diversity were additive rather than interdependent.
Facilitation is an important driver of woody species richness at macroecological scales because it added up to diversity in most sites, with enhanced contribution with increased stress. The HBM was not shaped by species interactions. Results suggest that facilitation may act in Mediterranean vegetation buffering against critical transitions between states allowing woody plant communities to cope with the rise in aridity expected with global warming.

Introduction Understanding the factors that shape species diversity at local, regional and global scales is a major challenge in ecology (Ricklefs, 2004; Harrison & Cornell, 2008; Allesina & Levine, 2011). Species interactions are considered determinants of community diversity, with competition and predation thought to play leading roles in community organization (e.g. Menge & Sutherland, 1976; Tilman, 1982). Likewise, productivity, which is strongly linked with environmental/climatic factors, likely determines macroecological and regional gradients in species richness (e.g. Grime, 1973; Abrams, 1995). Grime’s humpedback model (HBM) of species richness (Grime, 1973) was an important conceptual advance that merged the effects of productivity (or environmental stress) and species interactions (primarily through competition) on species diversity. According to this hypothesis, species richness peaks at intermediate levels along productivity–stress gradients, with competitive species dominating towards the high productivity extreme, stress-tolerant species dominating towards the stressful extreme and both strategies coexisting at intermediate levels. Grime’s influential model continues to be evaluated, with persistent debate and refinements based on studies with animal and plant communities (Mittelbach 464 New Phytologist (2016) 211: 464–476 www.newphytologist.com

et al., 2001; Michalet et al., 2006; Adler et al., 2011; Fraser et al., 2015). However, in the last two decades, studies on the influence of plant–plant interactions on community structure, assembly and diversity have moved the focus from competition towards facilitation in stressful environments (Bertness & Callaway, 1994; Bruno et al., 2003; Valiente-Banuet et al., 2006; Verd
u et al., 2009). Since Bertness & Callaway (1994) formulated the stress gradient hypothesis (SGH), information on the importance of facilitative interactions for plant performance in stressful environments, such as alpine, arid or salt marsh ecosystems, has accumulated rapidly (Brooker et al., 2008; Soliveres et al., 2015); however, consensus is far from being reached on the generality and the exact form (whether a monotonic increase or unimodal) of the relationship between the frequency (and intensity) of positive interactions and the level of the environmental stress (Maestre et al., 2005; Lortie & Callaway, 2006; He et al., 2013; Michalet et al., 2014; Soliveres & Maestre, 2014). Water availability is one of the most general limiting resources for plant productivity worldwide, and aridity is among the most important stresses influencing plant community diversity and productivity and ecosystem functioning (Maestre et al., 2012). In particular, aridity is a widespread stress in the Mediterranean region and the one influencing species interaction and plant Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist community composition and structure throughout the region most obviously (Armas et al., 2011; Pugnaire et al., 2011; Gross et al., 2013). In this study, we examine how the balance between facilitation and competition contributes to the species diversity of woody plant communities over most of the productivity–aridity gradient of the Western Mediterranean floristic region (a plant biodiversity hotspot; M
edail & Qu
ezel, 1997). We do this by assessing both the HBM and the SGH, two influential hypotheses that more commonly have been examined independently in previous studies. Studies connecting the HBM and the SGH have been largely theoretical (Hacker & Gaines, 1997; Michalet et al., 2006). Moreover, there have been few attempts to evaluate empirically how the competition–facilitation balance affects species diversity at the community level across stress gradients (Soliveres et al., 2011, 2012), with even fewer studies at macroecological scales (Cavieres et al., 2014). One clear connection between HBM and SGH emerges when productivity is limited primarily by aridity, a condition typical in Mediterranean ecosystems. In this case, both the balance between competition and facilitation and species richness theoretically will covary with aridity. Therefore, unlike recent studies that examined the contribution of facilitation to the macroecological patterns of species diversity (e.g. Cavieres et al., 2014) but did not explicitly address such hypotheses, we tested both hypotheses and examined whether the stress gradient and the plant–plant interaction balance have synergistic (i.e. interactive) or independent (i.e. additive) effects on species diversity. Merging both hypotheses answers the recent calls for efforts to understand the multivariate drivers of species richness (Adler et al., 2011; Fraser et al., 2015). It is notable that the HBM and further modifications of this model to incorporate facilitation (e.g. Michalet et al., 2006) mechanistically postulate that the influence of species interactions on community species richness across productivity–stress gradients acquires different weights depending on the position on the gradient. Thus, we should expect synergistic effects of the balance of interactions and productivity/aridity on species richness. By contrast, the independent effects of both drivers would imply that the competition–facilitation balance affects species richness over the entire gradient in a relatively comparable manner and that the humped-back distribution, when it occurs, is not shaped by species interactions but rather by some alternative mechanism (see later). Crossing over boundaries of vegetation types is important to detect macroecological hump-shaped patterns of species richness across productivity–stress gradients for vascular plants (Mittelbach et al., 2001) and to envisage the mechanisms involved in such patterns; however, few studies have been conducted at a sufficient scale to be conclusive (Fraser et al., 2015). Moreover, most insights in this direction have come from meta-analyses rather than from studies conducted simultaneously at many sites and at different scales using homogenous sampling protocols (but see Adler et al., 2011; Fraser et al., 2015). Across the productivity– aridity gradient of the Mediterranean region, a transitional shift from forest to scrub communities typically occurs, and the different dominant lifeforms and physiognomies of these communities may affect both the balance of the interaction between plant Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

Research 465

species and species richness. Furthermore, the western range of the floristic Mediterranean region extends to Mediterranean Europe, northwestern Africa, and the Canary Islands (Qu
ezel, 1985; M
edail & Qu
ezel, 1997), and the contemporary floras of these three areas differ in regional species pools, biogeographical origins and turnovers of species (M
edail & Qu
ezel, 1999; Lavergne et al., 2013). Modelling species richness in the overall context of the region may confound some of these issues that influence diversity. Therefore, as another goal of this study, we explored whether the HBM and the SGH are maintained independently of geographic scale, vegetation type (forest vs scrublands) and/or biogeographical region. To summarize, the wide productivity–aridity gradient of the Mediterranean region and its complex biogeographic history make it an ideal subject to assess the occurrence of diversity–productivity relationships and examine the mechanisms, in addition to species interactions, that may be involved in such a pattern. Linking the HBM and SGH at macroecological scales can provide new insights with which to understand gradients of species diversity; however, the studies conducted to date have lacked a clear macroecological context. By our empirical combination of both hypotheses, we also provide novel insights enhancing understanding of the consequences of global warming on the maintenance of plant communities and species diversity (Soliveres et al., 2015), a concern particularly relevant in the Mediterranean Basin.

Materials and Methods Study sites and vegetation types We sampled 54 study zones belonging to 16 woody plant communities (one to five zones per community, most commonly four), from mesic forests to arid scrublands in Spain (25 zones), Morocco (21) and the Canary islands (8), spanning a geographical range of c. 1750 km. The study zones were occupied by mature communities typical of the environmental conditions of the study area, and, when possible, the study sites were within protected and/or low-impact areas to minimize anthropogenic effects on recruitment. Largely disturbed sites were excluded because land-use history and management change the form of the regional- and global-scale relationships of species richness and productivity (Adler et al., 2011), inflating the chance of a humped-back relationship. We studied a representative sample of the forest and scrubland types of the study region. Data and locations of the study zones and communities surveyed in this work are provided in Supporting Information Table S1 and Fig. 1. Based on the dominant life-forms, 32 of these zones were considered forests and 22 scrublands (Fig. 1; Table S1). Sampling design The surveys were conducted in five 65 9 50 m plots in each study zone, with the plots separated by > 100 m and arranged in parallel lines, topography permitting. Plots were planned to cross relatively heterogeneous canopy coverage to allow for the open New Phytologist (2016) 211: 464–476 www.newphytologist.com

466 Research

New Phytologist

Fig. 1 Study site locations and major plant communities considered in this study. Plant communities considered forests were Pinus canariensis forests, Cedrus atlatica forests, Argania spinosa forests, several types of mixed Pine–oak forests (Pinus halepensis–Q. faginea; P. nigra–Q. pyrenaica, P. pinaster– Q. faginea, P. halepensis–Q. ilex), Abies pinsapo forests, Quercus ilex and Quercus suber oak forests. Plants communities considered scrublands were: Morocco and Canary Islands Euphorbia scrublands, Juniperus phoenicea scrublands, Tetraclinis articulata scrublands, Maytenus senegalensis scrublands and Ziziphus lotus scrublands. Numbering of study site localities follow that given in Supporting Information Table S1, where further details of each locality is provided. The inset represents the Mediterranean floristic ecoregion (ecoregion of Mediterranean forest, woodlands and scrublands. Global terrestrial Ecoregions, WWF). GAI, global aridity index.

interspaces and small gaps that are generated by natural disturbances (e.g. tree falls and wild ungulates) or patches of shallow soil. Within each of these plots, surveys of juveniles were conducted using paired samples, with each pair composed of the ground under the canopy of an adult of the most common plant species in the community (here considered the nurses) and an adjacent open of similar area outside the cover of that adult. The cover of the adult of each pair was assessed with the measurement of two perpendicular diameters of the canopy. Paired sampling of established juveniles has been used in previous studies to assess variation in the intensity of facilitative interactions in relation to climatic variation and/or or stress gradients (e.g. Soliveres et al., 2011; Cavieres et al., 2014). Moreover, the use of standing regeneration niches is common in studies that evaluate facilitation, test the stress-gradient hypothesis (SGH) and/or explore the contribution of facilitation to species richness at the community level (Valiente-Banuet et al., 2006; Soliveres et al., 2011, 2012; Cavieres et al., 2014). The cover of each plant species was estimated in the 65 9 50 m plots along a 65 9 5 m transect centred in the plot. Species interaction balance estimation Advocates of the SGH have noted several uncertainties that potentially alter its verification, such as misdefined or incomplete stress gradients, confounding effects of multiple stresses, New Phytologist (2016) 211: 464–476 www.newphytologist.com

incomplete or disparate datasets for meta-analytical tests and use of barely stress-sensitive life stages (He et al., 2013; He & Bertness, 2014). We attempted to control some of these uncertainties: our study encompassed nearly the entire productivity–aridity gradient in the Mediterranean region from a value of 0.1 in Abies pinsapo forests to semidesertic areas with aridity up to 0.9; we avoided alpine scrublands with freezing temperatures to prevent a confounding stress; we defined clearly that the scale of examination of the SGH was at the community level of response rather than at a species-pairwise level (Soliveres et al., 2015); we focused on pre-reproductive life-stages, which are preferred to adults for tests of the SGH; and our data were all obtained with the identical sampling protocol. The estimate of the interaction balance was based on the comparison of established juveniles (i.e. saplings, not seedlings) beneath adult plants vs open. To estimate the result of the interaction between species, we used the relative interaction index, RII (Armas et al., 2004), which is frequently used to characterize variation in facilitation across stress gradients (Armas et al., 2011; Soliveres et al., 2011; Cavieres et al., 2014). The index has values between –1 (competition) and +1 (facilitation) and is based on the performance of a species when growing alone or when interacting with other species. Here, RII was calculated for each recruiting species using the following equation: RII = (juveniles beneath nurses  juveniles in open)/(juveniles beneath nurses + juveniles in open). Once the RII for each species in a Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist community was obtained, values were averaged across species to obtain a community-level RII, which was used in the analyses. The RII was calculated only for those species with at least five juveniles in the samples of the juvenile bank. Species richness estimation Congruent with the estimation of the interaction balance, species richness here referred only to juveniles (not seedlings) of woody species (trees and scrubs) from each study zone. Caution was used in determining juveniles of resprouting species, and some species with clonal and rhizomatous growth (e.g. Rubus, Smilax, Jasminum, Osyris, Clematis, Rubia) were not included in the analyses. Restricting the monitoring of species richness to woody species was relevant because, first, it focused on relatively similar growth forms and therefore richness estimates were directly comparable and, second, the counts were, to some extent, independent of yearly variation in production. Despite the similar sampling effort within each zone, the number of juveniles detected varied greatly among communities and zones. Thus, we used rarefaction to estimate juvenile species richness, and adopted the projections of species richness from rarefaction curves using ESTIMATES v.9.0 (Colwell, 2013; see Fig. S1 for some examples on estimated rarefaction curves). For each zone, species richness from rarefaction curves was standardized to 200 individuals (S200). We complemented this estimation with a species richness estimator that accounted for rarity (Chao1). Both estimates were strongly correlated across zones (r = 0.87) and with the total number of species detected as juveniles (r = 0.90 for S200; r = 0.94 for Chao1). The estimates of S200 and Chao1 were based on the number of juveniles detected rather than on the number of samples (or area sampled) because each sample of surveyed juveniles was from a very different area and depended on the cover area of the adult specimen and the species lifeform. The use of individual-based rarefaction curves to estimate species richness is common in ecology (Gotelli & Colwell, 2001; Gotelli & Chao, 2013) and is particularly convenient to control for sampling efforts when very different total numbers of individuals are sampled in each community. Contribution of facilitation to species richness Following Cavieres et al. (2014), we evaluated the magnitude of the contribution of facilitation to species richness by estimating the increase in species richness index (ISR), which measures the proportion of the species in the community that do not recruit in the open (i.e. obligate facilitated). This index was calculated by obtaining total juvenile species richness and juvenile species richness in open by rarefaction. ISR = (STOTAL – SOPEN)/STOTAL, where S is the species richness. Aridity We used the global aridity index (GAI) of CGIAR-CSI (http:// www.cgiar-csi.org; Zomer et al., 2007, 2008) to the resolution of 30 arc s (c. 1 km at the equator). This index combines mean Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

Research 467

annual precipitation and potential evapotranspiration. With an increase in this index, an increase in humidity in the environment is indicated; therefore, our analyses and graphs were based in 1-GAI for a straightforward interpretation. Climatic indices calculated both on an annual and a seasonal basis are relevant for vegetation and, in particular, for estimating climatic stress and the influence on the balance of interactions (Soliveres & Maestre, 2014). We used annual climatic data rather than seasonal data because annual-based climatic indices are commonly used in studies on facilitative interactions that consider rainfall, temperature or aridity as the primary stressors (e.g. Soliveres et al., 2012) and, more commonly, are better descriptors of the relationships between facilitative interactions and climatic stress than seasonal variables (see recent review of Soliveres & Maestre, 2014). Because the humped-back model (HBM) is formulated as a species diversity–productivity relationship, we also used indirect estimates of productivity to demonstrate that the aridity gradient was inversely related to productivity. We estimated productivity (net primary production, NPP) in each plot using remote sensing images from MODIS (Running et al., 2015) at a resolution of 200M 9 200M pixels (see Gait
an et al., 2014 for a similar approach). Furthermore, we used the sum of the woody plant cover of the different vegetation layers as a surrogate for productivity. Statistical analyses General linear models (GLMs) were used to analyse the variation among regions and between vegetation types (both treated as fixed effects) for species richness, RII and ISR. Because of the nonlinear relationships proposed by the HBM and the SGH, these hypotheses were first assessed independently for the overall dataset using nonlinear quadratic regressions (e.g. Mittelbach et al., 2001; Maestre & Cortina, 2004). The response variables in these regressions were species richness and RII for the HBM and SGH, respectively, whereas the predictors were simple and quadratic terms of 1-GA (aridity) in the tests of both models. Quadratic regressions were conducted using GLM models in STATISTICA v.8 (StatSoft Inc., 2007). Because the estimators of species richness (S200 and Chao1) are typically continuous rather than counts, and the RII is typically normally distributed (Armas et al., 2004), we adopted normal error and identity link functions in our tests of the HBM and SGH. The HBM and SGH were subsequently assessed within each region and for each vegetation type independently. The possibility of synergistic effects of aridity and interaction balance on species richness was tested using surface regression. Surface regression is a type of GLM that includes linear, quadratic and interaction terms between two continuous predictors (X and Y) on a single response variable (Z): Z ¼ a þ b1 X þ b2 Y þ b3 X 2 þ b4 Y 2 þ b5 XY Thus, it permits the modelling of the independent effects of each predictor (whether linear or quadratic) on the response variable and also the possibility of interdependence (i.e. that the New Phytologist (2016) 211: 464–476 www.newphytologist.com

New Phytologist

468 Research

effects of both predictors covary or depend on one another; the XY term in the equation earlier). It is particularly useful when the combination of continuous variables is likely to affect a response variable in complex ways, for example, nonlinearly, that is, giving rise to a surface that depicts the combination of values of the predictors that renders maximum and minimum in the response variable. Interpretation of the effects of the surface regression is similar to that for other factorial GLMs (e.g. ANOVA, ANCOVA). When no significant interaction of the predictors occurs (i.e. the XY term is not significant), the effects are additive (when both predictors have significant effects, linear or quadratic, on the response variable). By contrast, when the interaction of the predictors is significant, the effects of the predictors are interpreted as synergistic (i.e. dependent on one another). Surface regression is an appropriate statistical technique to merge the SGH and HBM analytically because productivity–aridity is the common axis (predictor) of the relationships depicted by these hypotheses. Moreover, both productivity–aridity and facilitation influence species diversity, but their link to explain species richness is unresolved. Therefore, with surface regression, we tested for linear and quadratic relationships of species richness to productivity–stress (as is typically conducted in HBM tests), linear or quadratic relationships of species richness to the relative interaction index (i.e. the contribution of facilitation–competition balance to species richness), and whether the effects of productivity–aridity and facilitation–competition on species richness covaried (as implicit in the mechanistic basis formulated by Grime for the HBM and further modifications; Michalet et al., 2006). Because species richness and/or RII could vary among regions and types of vegetation (forests and scrublands), we additionally tested the HBM, SGH and the surface regression by incorporating region and vegetation type separately as blocking effects in the overall models. Moreover, aridity also influences other variables that can affect productivity and the outcome of plant–plant interactions (e.g. soil nutrients and effects of human activities such as grazing; e.g. Ulrich et al., 2014). Spatial covariation of aridity and these environmental variables along a geographical gradient may result in confounding effects. To avoid this

possibility, we repeated all analyses of the SGH, the HBM and the surface regression with and without geographic coordinates (longitude and latitude) as covariables. Incorporation of spatial coordinates did not change the statistical significance of any linear or quadratic coefficient. Therefore, in favor of simplicity, we opted to maintain the analyses, tables, figures and results without the geographic coordinates. Tests with geographic coordinates are provided in Table S2. The assumptions of the statistical tests used were met without transformation of the original variables.

Results In total, we recorded 26 844 juveniles of woody plants (mean across zones = 497.1; median = 278.5) belonging to 206 species, 102 genera and 42 families (Table S3). The mean observed juvenile species richness per zone was 13.9 (range 5–28; median = 13). Some species in the community were not observed as juveniles. Thus, the average total richness per zone was 17.2 (range 6–48; median = 15), with the average juvenile species richness representing 80.8% of the total. The total species richness and juvenile species richness were strongly correlated (r = 0.90; P 0.75 (Table S1). Consistent with this idea the ISR was lower in the Canary Islands and southern Spain than in Morocco, where the aridity expands to the most severe values. Likewise, Cavieres et al. (2014) found that the increase in species Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

Research 473

richness by facilitation declines with productivity in alpine communities at a global scale. Assuming a unimodal relationship between frequency (or intensity) of positive interactions and stress, with a collapse of facilitation at the extremes of the gradient, Michalet et al. (2006) hypothesized that facilitation would contribute to species richness with maximum scores at intermediate levels of stress. Our results did not confirm this hypothesis. We found increasing ISR with aridity which, indeed, could be expected from the incorporation into the species diversity curve of the monotonic increase in facilitation postulated by the SGH. The relevance of facilitation for species richness was also manifested after controlling for the interdependence of species richness with the productivity–aridity gradient. According to our surface regression models, RII might positively affect species diversity independent of the level of aridity, which means that benefits other than an amelioration of aridity stress are also achieved for the recruits that grow under adults in the local communities studied. These benefits include protection against herbivory, provision of soil nutrients and reduced interspecific belowground competition for shared mycorrhizal networks (Moora & Zobel, 2010; Pugnaire et al., 2011; Soliveres et al., 2012). Notably, our models also suggest increased richness at highly negative RII values (Fig. S2), which might reflect the occurrence of species strictly recruiting in the open (nurse avoiders) and also account for some additional richness through niche differentiation mediated by functional trait divergence (Gross et al., 2013). This situation could occur with the shade-intolerant species in Mediterranean forests that require recurrent disturbance to persist (Alc
antara et al., 2015). In short, facilitation was an important driver of woody species richness at a macroecological scale because it added up to diversity in most sites and the contribution to diversity increased with aridity. Does the balance of interactions shape the humped-back diversity–aridity relationship? As stated in the introduction, if the species interaction balance shaped the relationship between species richness and productivity–aridity, we would expect interdependent effects of aridity and RII on species richness. However, we found that the humped-back relationship between species richness and aridity was maintained in Mediterranean woody plant communities at a continental scale and did not depend on how the interaction balance varied across the aridity gradient. Thus, other drivers are responsible for the humpedback relationship. In a recent review, Michalet & Touzard (2010) proposed that the relative importance of the primary ecological filters of plant diversity depends on the geographic scale. They suggested that at regional (< 1000 km) and continental (> 1000 km) scales the patterns of community richness are driven primarily by stress and chance biogeographical events (Huston, 1999; Lortie et al., 2004). Our continental humped-back pattern might be consistent with this proposal. On the one hand, the complex biogeographical history of the Mediterranean region has repeatedly melded species pools (sensu Cornell & Harrison, 2014) and lineages of different biogeographical origins (i.e. Eurosiberian, Mediterranean, Irano-Turanian and SaharoNew Phytologist (2016) 211: 464–476 www.newphytologist.com

New Phytologist

474 Research

Arabian; Thompson, 2005), that would be particularly pronounced at intermediate positions of the gradient. On the other hand, parallel to and to a large extent shaped by the aridity gradient of the Mediterranean region, there is a transition from typical temperate forest to savanna-like woodlands and finally to arid scrublands. We propose that the turnover of these vegetation formations inherently shapes the unimodal relationship of species richness with aridity. Supporting this proposal, our analysis of the HBM by vegetation type showed that the overall humpedback relationship was depicted as a trend of species richness to increase with aridity in forests and decrease in scrublands, which is linked with the replacement of lifeforms (trees to scrubs) along the aridity gradient (Fig. 3). We hypothesize that vegetation type turnover involves a concomitant transition of predominant plant functional traits and of the level in which functional traits are filtered by environment (Gross et al., 2013). With these two nonexclusive explanations, we would expect that the patterns of functional, biogeographical or phylogenetic diversity would reflect the species diversity pattern across the aridity gradient; the evaluation of these possibilities is currently under investigation. Concluding remarks: implications for climate change concerns Our study is based on spatial correlations and therefore some caution is needed to interpret our results. Although HBM can be strongly supported with spatial patterns, further experimental evidence is needed in case of SGH tests. Keeping this in mind, our results provide insights to improve our understanding of the consequences of global warming for woody plant communities and the maintenance of species diversity in the Western Mediterranean floristic region. Drylands are ecosystems prone to critical transitions from vegetated to nonvegetated stable states under global change drivers such as climate change and intensified land use (K
efi et al., 2007a; Verwijmeren et al., 2013). In fact, the humped-back pattern of diversity–aridity relationship, that was detected at a continental scale and shaped across community boundaries, suggests that the Western Mediterranean floristic region may undergo critical transitions from forest to scrubland and from scrubland to nonvegetated states with an increase in aridity. Recent connections between critical transition theory and facilitation have led to the proposal that declines in facilitation with stress would increase the chances of critical transitions from vegetated to nonvegetated stable states in arid ecosystems (Verwijmeren et al., 2013; and references therein). We propose that when the relationship of facilitation and aridity is unimodal then increased drought in the region resulting from climate change would cause a decline in the frequency and intensity of facilitative interactions in communities located at present on the high aridity side of the gradient, which could potentially trigger the critical transition to nonvegetated states. By contrast, our finding of a monotonic increase of facilitation intensity with aridity would suggest that the Mediterranean woody vegetation is, to some extent, resistant to critical transitions from forest to scrubland and to desert, because facilitation would act as a buffering New Phytologist (2016) 211: 464–476 www.newphytologist.com

mechanism against such transitions. Additionally, with a monotonic increase, the persistence of species would increase and, consequently, maintain the ecosystem multifunctionality of drylands through species diversity (Maestre et al., 2012). It has been suggested that the intensity of facilitation may determine the position of a critical transition threshold (K
efi et al., 2007b) and serve as a warning signal in the forecasting of transition. Our study illustrates that a relatively rapid and repeatable protocol of juvenile monitoring produces indices of facilitation intensity and species richness that are useful to such ends.

Acknowledgements We are most grateful to Jes
us Bastida, Inma Cancio, Manuel Pulgar and Carlos Ruiz for their invaluable help with fieldwork. The Spanish Ministerio de Ciencia e Innovaci
on and FEDER of the European Commission funded this work with projects CGL2009-08130 and CGL2012-36776.

Author contributions P.J.R. developed the conceptual framework of this study; P.J.R., J.M.A. and A.J.M. designed the sampling protocols; P.J.R. and J.M.A. analysed the data; P.J.R., J.M.A., A.J.M. and A.M.S-L. participated in data collection, processing and interpretation; and P.J.R., J.M.A. and A.J.M. wrote the paper.

References Abrams P. 1995. Monotonic or unimodal diversity–productivity gradients: what does competition theory predict? Ecology 76: 2019–2027. Adler PB, Seabloom EW, Borer ET, Hillebrand H, Hautier Y, Hector A, Harpole WS, O’Halloran LR, Grace JB, Anderson TM et al. 2011. Productivity is a poor predictor of plant species richness. Science 333: 1750–1753. Alc
antara JM, Rey PJ, Manzaneda JM. 2015. A model of plant community dynamics based on replacement networks. Journal of Vegetation Science 26: 524–537. Allesina S, Levine JM. 2011. A competitive network theory of species diversity. Proceeding of the National Academy of Sciences, USA 108: 5638–5642. Armas C, Ordiales R, Pugnaire FI. 2004. Measuring plant interactions: a new comparative index. Ecology 85: 2682–2686. Armas C, Rodr
ıguez-Echeverr
ıa S, Pugnaire FI. 2011. A field test of the stressgradient hypothesis along an aridity gradient. Journal of Vegetation Science 22: 818–827. Bertness MD, Callaway RM. 1994. Positive interactions in communities. Trends in Ecology & Evolution 9: 191–193. Brooker RW, Maestre FT, Callaway RM, Lortie CJ, Cavieres LA, Kunstler G, Liancourt Pierre, Tielb€orger K, Travis JMJ, Anthelme F et al. 2008. Facilitation in plant communities: the past, the present, and the future. Journal of Ecology 96: 18–34. Bruno JF, Stachowicz JJ, Bertness MD. 2003. Inclusion of facilitation into ecological theory. Trends in Ecology & Evolution 18: 119–125. Cavieres LA, Badano I. 2010. Consequences of facilitation on species diversity in terrestrial plant communities. In: Pugnaire FI, ed. Positive plant interactions and community dynamics. Boca Raton, FL, USA: CRC Press/Taylor & Francis, 39–57. Cavieres LA, Brooker RW, Butterfield BJ, Cook BJ, Kikvidze Z, Lortie CJ, Michalet R, Pugnaire FI, Sch€ob C, Xiao S et al. 2014. Facilitative plant interactions and climate simultaneously drive alpine plant diversity. Ecology Letters 17: 193–202. Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist Colwell RK. 2013. EstimateS: Statistical estimation of species richness and shared species from samples, v.9. URL http://purl.oclc.org/estimates (accessed June 2013). Cornell HV, Harrison SP. 2014. What are species pools and where are they important. Annual Review of Ecology, Evolution and Systematics 45: 45–67. Fraser LH, Pither J, Jentsch A, Sternberg M, Zobel M, Askarizadeh D, Bartha S, Beierkuhnlein C, Bennett JA, Bittel A et al. 2015. Worldwide evidence of a unimodal relationship between productivity and plant species richness. Science 349: 302–315. Gait
a n JJ, Bran D, Oliva G, Maestre FT, Aguiar MR, Jobb
agy EG, Buono GG, Ferrante D, Nakamatsu VB, Ciariet G et al. 2014. Vegetation structure is as important as climate to explain ecosystem function across Patagonian rangelands. Journal of Ecology 102: 1419–1428. G
omez-Aparicio L, Zamora R, G
omez JM, H
odar JA, Castro J, Baraza E. 2004. Applying plant facilitation to forest restoration: a meta-analysis of the use of shrubs as nurse plants. Ecological Applications 14: 1128–1138. Gotelli NJ, Chao A. 2013. Measuring and estimating species richness, species diversity, and biotic similarity from sampling data. In: Levin SA, ed. Encyclopedia of biodiversity, 2nd edn, vol. 5. Waltham, MA, USA: Academic Press, 195–211. Gotelli NJ, Colwell RK. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4: 379– 391. Grime JP. 1973. Competitive exclusion in herbaceous vegetation. Nature 242: 344–347. Gross N, B€orger L, Soriano-Morales SI, Le Bagousse-Pinguet Y, Quero JL, Garc
ıa-G
omez M, Valencia-G
omez E, Maestre FT. 2013. Uncovering multiscale effects of aridity and biotic interactions on the functional structure of Mediterranean shrublands. Journal of Ecology 101: 637–649. Hacker SD, Gaines SD. 1997. Some implications of direct positive interactions for community species diversity. Ecology 78: 1990–2003. Harrison SP, Cornell HV. 2008. Toward a better understanding of regional causes of local species richness. Ecology Letters 11: 969–979. He Q, Bertness MD. 2014. Extreme stresses, niches, and positive species interactions along stress gradients. Ecology 95: 1437–1443. He Q, Bertness MD, Altieri A. 2013. Global shifts towards positive species interactions with increasing environmental stress. Ecology Letters 16: 695–706. Holmgren M, Scheffer M. 2010. Strong facilitation in mild environments: the stress gradient hypothesis revisited. Journal of Ecology 98: 1269–1275. Holzapfel C, Tielb€orger K, Parag HA, Kigel J, Sternberg M. 2006. Annual plant–shrub interactions along an aridity gradient. Basic and Applied Ecology 7: 268–279. Huston MA. 1999. Local processes and regional processes: appropriate scales for understanding variation in the diversity of plants and animals. Oikos 86: 393–401. K
efi S, Rietkerk M, Alados CL, Pueyo Y, Papanastasis VP, Elaich A, de Ruiter PC. 2007a. Spatial vegetation patterns and imminent desertification in Mediterranean arid ecosystems. Nature 449: 213–217. K
efi S, Rietkerk M, van Baalen M, Loreau M. 2007b. Local facilitation, bistability and transitions in arid ecosystems. Theoretical Population Biology 71: 367–379. Lavergne SA, Hampe A, Arroyo J. 2013. In and out of Africa: how did the Strait of Gibraltar affect plant species migration and local diversification. Journal of Biogeography 40: 24–36. Lortie CJ, Brooker RW, Choler P, Kikvidze Z, Michalet R, Pugnaire FI, Callaway RM. 2004. Rethinking plant community theory. Oikos 107: 433–438. Lortie CJ, Callaway RM. 2006. Re-analysis of metaanalysis: support for the stress-gradient hypothesis. Journal of Ecology 94: 7–16. Maestre FT, Cortina J. 2004. Do positive interactions increase with abiotic stress? A test from a semi-arid steppe. Proceedings of the Royal Society B 271: S331–S333. Maestre FT, Quero JL, Gotelli NJ, Escudero A, Ochoa V, Delgado-Baquerizo M, Garc
ıa-G
omez M, Bowker MA, Soliveres S, Escolar C et al. 2012. Plant species richness and ecosystem multifunctionality in global drylands. Science 335: 214–218. Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

Research 475 Maestre FT, Valladares F, Reynolds JF. 2005. Is the change of plant–plant interactions with abiotic stress predictable? A meta-analysis of field results in arid environments. Journal of Ecology 93: 748–757. M
edail F, Qu
ezel P. 1997. Hot-spot analysis for conservation of plant biodiversity in the Mediterranean Basin. Annals of the Missouri Botanical Garden 84: 112–127. M
edail F, Qu
ezel P. 1999. The phytogeographical significance of S.W. Morocco compared to the Canary Islands. Plant Ecology 140: 221–244. Menge BA, Sutherland JP. 1976. Species diversity gradients: synthesis of the roles of predation, competition, and temporal heterogeneity. The American Naturalist 110: 351–369. Michalet R. 2006. Is facilitation in arid environments the result of direct or complex interactions? New Phytologist 169: 3–6. Michalet R, Bagousse-Pinguet YL, Maalouf JP, Lortie CJ. 2014. Two alternatives to the stress-gradient hypothesis at the edge of life: the collapse of facilitation and the switch from facilitation to competition. Journal of Vegetation Science 25: 609–613. Michalet R, Brooker RW, Cavieres LA, Kikvidze Z, Lortie CJ, Pugnaire FI, Valiente-Banuet A, Callaway RM. 2006. Do biotic interactions shape both sides of the humped-back model of species richness in plant communities? Ecology Letters 9: 767–773. Michalet R, Brooker RW, Lortie CJ, Maalouf JP, Pugnaire FI. 2015. Disentangling direct and indirect effects of a legume shrub on its understorey community. Oikos 124: 1251–1262. Michalet R, Touzard B. 2010. Biotic interactions, biodiversity and community productivity. In: Pugnaire FI, ed. Positive plant interactions and community dynamics. Boca Raton, FL, USA: CRC Press/ Taylor & Francis Group, 59– 78. Mittelbach GG, Steiner CF, Scheiner SM, Gross KL, Reynolds HL, Waide MR, Willig MR, Dodson SI, Gough L. 2001. What is the observed relationship between species richness and productivity? Ecology 82: 2381–2396. Molina-Vanegas R, Aparicio A, Pina FJ, Vald
es B, Arroyo J. 2013. Disentangling environmental correlates of vascular plant biodiversity in a Mediterranean hotspot. Ecology and Evolution 3: 3879–3894. Moora M, Zobel M. 2010. Arbuscular mycorrhizae and plant–plant interactions. Impact of invisible world on visible patterns. In: Pugnaire FI, ed. Positive plant interactions and community dynamics. Boca Raton, FL, USA: CRC Press/ Taylor & Francis Group, 79–98. Pugnaire FI, Armas C, Maestre FT. 2011. Positive plant interactions in the Iberian Southeast: mechanisms, environmental gradients, and ecosystem function. Journal of Arid Environments 75: 1310–1320. Qu
ezel P. 1985. Definition of the Mediterranean regi
on and the origin of its flora. In: G
omez-Campo C, ed. Plant conservation in the Mediterranean area. Dordrecht, the Netherlands: Dr W. Junk, 9–24. Ricklefs RE. 2004. A comprehensive framework for global patterns in biodiversity. Ecology Letters 7: 1–15. Running S, Mu Q, Zhao M. 2015. MOD17A3H MODIS/Terra Net Primary Production Yearly L4 Global 500 m SIN Grid V006. NASA EOSDIS Land Processes DAAC. doi: 10.5067/MODIS/MOD17A3H.006. Siles G, Rey PJ, Alc
a ntara JM. 2010. Post-fire restoration of Mediterranean forests: testing assembly rules mediated by facilitation. Basic and Applied Ecology 11: 422–431. Soliveres S, Eldridge DJ, Hemmings F, Maestre FT. 2012. Nurse plant effects on plant species richness in drylands: the role of grazing, rainfall and species specificity. Perspectives in Plant Ecology, Evolution and Systematics 14: 402–410. Soliveres S, Eldridge DJ, Maestre FT, Bowker MA, Tighe M, Escudero A. 2011. Microhabitat amelioration and reduced competition among understory plants as drivers of facilitation across environmental gradients: towards a unifying framework. Perspectives in Plant Ecology, Evolution and Systematics 13: 247–258. Soliveres S, Maestre FT. 2014. Plant–plant interactions, environmental gradients and plant diversity: a global synthesis of community-level studies. Perspectives in Plant Ecology, Evolution and Systematics 16: 154–163. Soliveres S, Smit C, Maestre FT. 2015. Moving forward on facilitation research: response to changing environments and effects on the diversity, functioning and evolution of plant communities. Biological Reviews 90: 297–313. New Phytologist (2016) 211: 464–476 www.newphytologist.com

New Phytologist

476 Research StatSoft Inc. 2007. STATISTICA (data analysis software system), v.8.0. [WWW document] URL http://www.statsoft.com [accessed July 2007]. Thompson JD. 2005. Plant evolution in the Mediterranean. New York, NY, USA: Oxford University Press. Tilman D. 1982. Resource competition and community structure. Princeton, NJ, USA: Princeton University Press. Ulrich W, Soliveres S, Maestre FT, Torices R, Gotelli NJ, Quero JL, DelgadoBaquerizo M, Ochoa V, Gozalo B, Valencia E et al. 2014. Climate and soil attributes determine plant species turnover in global drylands. Journal of Biogeography 41: 2307–2319. Valiente-Banuet A, Rumebe AV, Verd
u M, Callaway RM. 2006. Quaternary plant lineages sustain global diversity by facilitating tertiary lineages. Proceeding of the National Academy of Sciences, USA 103: 16812–16817. Verd
u M, Rey PJ, Alc
a ntara JM, Siles G, Valiente-Banuet A. 2009. Phylogenetic signatures of facilitation and competition in successional communities. Journal of Ecology 97: 1171–1180. Verwijmeren M, Rietkerk M, Wassen MJ, Smit C. 2013. Interspecific facilitation and critical transitions in arid ecosystems. Oikos 122: 341–347. Whittaker RJ, Heegaard E. 2003. What is the observed relationship between species richness and productivity? Ecology 84: 3384–3390. Zomer RJ, Bossio DA, Trabucco A, Yuanjie L, Gupta DC, Singh VP. 2007. Trees and water: smallholder agroforestry on irrigated lands in Northern India. IWMI Research Report 122. Colombo, Sri Lanka: International Water Management Institute. Zomer RJ, Trabucco A, Bossio DA, van Straaten O, Verchot LV. 2008. Climate change mitigation: a spatial analysis of global land suitability for clean development mechanism afforestation and reforestation. Agriculture, Ecosystems and Environment 126: 67–80.

Supporting Information Additional supporting information may be found in the online version of this article. Fig. S1 Estimated rarefaction curves for a sample of 12 study localities. Fig. S2 Contour plots of species richness in surface response regression models. Table S1 Compiled data in each study site Table S2 Complete set of tests of the humped-back model, the stress gradient hypothesis, and their combined analysis, incorporating latitude and longitude of the study sites as covariates Table S3 Complete list of plant species and families found as juveniles in this study Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews. Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication ‘as ready’ via Early View – our average time to decision is